The U.S. government has removed a licensing requirement that had prevented Anthropic from making its Mythos and Fable artificial intelligence models broadly available outside the United States, allowing the company to restore public access beginning July 1.
The decision reverses export restrictions introduced on June 12 that required Anthropic to obtain government approval before exporting the models to foreign users. Anthropic said complying with the licensing process at scale was not practical, prompting it to suspend public access to both models.
Government and Anthropic Reach Agreement
The policy change follows discussions between Anthropic and the U.S. Department of Commerce over security measures for advanced AI systems.
Secretary of Commerce Howard Lutnick said the company had agreed to strengthen its cooperation with the U.S. government.
“Anthropic has agreed to proactively detect and address security risks associated with the models; to work diligently with the U.S. government on protocols and standards and releases for Mythos, Fable and future models; and to inform the US government of any malicious activity.”
Anthropic had previously announced many of these commitments voluntarily before the export restrictions were introduced.
Pressure to Maintain U.S. AI Competitiveness
Mythos was initially released to a limited group of organizations in April to evaluate its ability to identify and exploit software vulnerabilities. A public version, Fable, followed in June with additional safety guardrails.
The restrictions drew criticism from some cybersecurity experts, who argued that existing voluntary safeguards already addressed many of the government’s stated concerns.
The decision to lift the licensing requirement also comes as AI developers in Asia continue to introduce increasingly capable foundation models, adding competitive pressure on U.S. policymakers to ensure American AI companies remain active in global markets.
Ongoing Regulatory Uncertainty
Before the latest decision, the Commerce Department had permitted Anthropic to provide Mythos only to a limited group of government-approved customers. Similar restrictions had also applied to OpenAI’s newest models.
The reversal highlights the evolving nature of U.S. AI policy as regulators seek to balance national security concerns with maintaining the global competitiveness of American AI developers.
Questions remain over the future regulatory framework for advanced AI models following recent policy shifts and a June executive order signaling greater federal oversight of future AI releases.